How much beta at 1mA 2MHZ for them MJE340 & 350? Will you be adding those BJTs to your models text soon?
I've got an HP3312A and a 400EL so I might try a few devices. I've got some real old devices here, 40409/10, and some MJE243/53 from the late 1970s, new old stock.
I have a PA amp out on loan that I'm fairly certain has RCA single diffused 2N3055s that would be interesting to test. I'm fairly certain that I also have an RCA 40251 that is in the 2N3055 family around here somewhere.
Last edited:
Hi Salas,
I went back and looked at my modeling notes and it turns out that at 1 mA I only measured AC beta on these devices up to 1 MHz. At 1 mA and 1 MHz, AC beta was 11.7 and 11.3 for the MJE340 and 350, respectively. This would correspond to approximately 5.9 and 5.6 at 2 MHz, respectively. As you know, ft decreases at low current, especially for a power device at only 1 mA, largely due to junction capacitances.
Yes, I'll be adding these devices to my set of models on my web page very soon.
Cheers,
Bob
Thanks Bob. I asked that because I use them at low current without dominant Miller cap some times and are nicely stable. The beta is also very near between NPN-PNP at low current as you found, that is nice. Do they keep up near at 10mA also? Will use your models as soon as you will make them available, because there is mist around the real characteristics of those classic handy BJTs.
Hi Andrew,
Measuring the ft of a transistor is really not that difficult, especially for devices with ft less than 100 MHz. In practice, one actually measures the ac current gain (ac beta) at a reasonably high frequency and then extrapolates ft by multiplying the current gain at the test frequency by the test frequency. If, for example, ac current gain is measured to be 12 at a test frequency of 1 MHz, then ft is estimated to be 12 MHz. That same transistor when measured at a test frequency of 500 kHz would probably show an ac beta of about 24.
It is important that the ac beta at the test frequency be already substantially attenuated from the ac beta at low frequencies so that the assumed 6 dB/octave roll-off in ac beta with frequency is valid. In general, the test frequency should be high enough so that the ac beta at the test frequency is no greater than 1/5 the low-frequency ac beta. It is also important that the test frequency not be so high that ac beta at the test frequency is less than about 5. These rules of thumb are not usually difficult to observe. Low-frequency ac beta can be conveniently found in the range of 1 kHz to 10 kHz.
A transistor with LF ac beta of 100, and having an ft of 100 MHz, will have an ac beta of about 100 at 1 MHz by straight Bode plot extrapolation, but in reality 1 MHz is the 3-dB frequency of ac beta, or f-beta, so ac beta will be down about 3 dB at 1 MHz, for an expected ac beta of 71 at 1 MHz. A test frequency of 1 MHz is inadequate for this transistor because the test frequency should be well above f-beta – another way of saying that ac beta should be less than 1/5 of low-frequency beta at the test frequency. These are not iron rules, however. If you test at only one octave above f-beta, you can expect about a 10% error in estimated ft when you multiply the test frequency by the ac beta at that frequency (the straight-line Bode approximation is off by about 1 dB one octave away from a pole frequency). If you are a bit cunning, you can estimate out some of this error, but then you need to know where f-beta is. If you need to use test frequencies substantially higher than 1 MHz, then more expensive test equipment may be needed.
Often, you will see datasheets for small-signal transistors that specify or plot ac beta at a test frequency of 20 MHz.
To make the measurement of ac beta, you typically will need a test oscillator and an ac voltmeter, both of which should be capable to 10 MHz. In some cases you may be able to get away with an audio generator an ac meter that go only to 1 MHz. I use an HP 652 test oscillator and an HP 400 EL ac voltmeter – both obtained on Ebay. You’ll also need two power supplies, one to supply collector voltage and one to supply and adjust base bias current through a large-value resistor.
My setup is simple. I operate the transistor in common-emitter mode with a 100 ohm collector load to a well-bypassed 10V supply (tweaked up a bit to obtain 10 V Vce after the drop across the 100 ohm collector resistor). The supply should be bypassed close to the transistor and its load with, say, a 10 uF electrolytic in parallel with a 0.1 uF ceramic. The low-value load resistance reduces an Miller effect on the measurement and also reduces the effects of test equipment capacitive loading.
Base bias current is supplied from the second supply through a large-value resistor like 100k or 1 Meg. Inject the test frequency into the base through a 100k resistor in series with a 0.1 uF capacitor.
Assume you have a transistor with dc beta of 100 that you will measure at Ic = 1 mA. It will need 10 uA of base current. Use a 1 Meg base bias resistor and feed it with about 10.7 V from the bias supply. Measure the dc drop across the 100-ohm load resistor and adjust it to 100 mV. Use a Hi-Z DC voltmeter isolated with 1k from the collector-side of the load resistor.
Measure LF ac beta by applying 1 kHz to the base through a 100k resistor at a voltage that will yield about 10 mV rms at the collector, as measured by your ac voltmeter. These numbers are not sacred; you just don’t want to get the measured level so high that distortion is significant (keep the transistor well away from clipping). You might choose to apply a fixed ac voltage of 100 mV rms, resulting in a base current of 1 uA rms. With LF ac beta of 100, this will create 100 uA ac collector current and 10 mV rms at the collector load. Taking the ratios of the base input resistor and collector load resistor (1000:1), and the input voltage against the output voltage (10:1), you’ll get a beta value of 100.
Now do the same thing at the higher test frequency, say 1 MHz. You can adjust the input voltage upward to get beck to the 10 mV rms neighborhood at the collector if you wish. Take those ratios to get ac HF beta and then multiply by the test frequency. This is all simpler than it sounds; just keep good measurement practices in mind and do sanity checks.
Cheers,
Bob
Hi Pete,
I bet they'll turn out to be pretty slow.
BTW, when I did the ft measurements I had the peak ac collector current of the device at about 10% of the dc bias current.
Cheers,
Bob
Hi Bob,
I know you stated way back that the devices in the RCA and Citation amp were probably very slow but I have to disagree with you. The 40409/10 are in the data book that I have
here and ft for both is 100 MHz - not bad at all. The 2N3055 and all related parts do not have ft specified in this RCA data book and I've seen people quote some very low numbers but I'm wondering if those are very much worst case even for the single diffused devices.
Last edited:
I think this is simpler, just do it at 5 - 10 MHz for audio parts:
MIL-M-38510/108A page 12?
http://www.dscc.dla.mil/Downloads/MilSpec/Docs/MIL-M-38510/mil38510ss108.pdf
Hi Salas,
These devices, like many, have higher ft and higher beta at 10mA as compared to 1mA.
I'm not sure what you mean by using them without dominant Miller cap. Is this in the VAS location? Are you relying on Miller effect from Ccb (which is quite nonlinear)?
Cheers,
Bob
I have actually seen some designs (and actual amps) that use the 340/350 do just that ..
I don't "trust" just Ccb. OS
Hi rolandong,
I'll see if I can dig that simulation out and post it here. Its been quite some time since I did that one. It may take a few days, as I'm still in the middle of trying to finish up some transistor models to post on my website. I'm also heading to Montreal this week for the FSI audio show.
I'm glad you like my book, and understand what you mean about there being fairly little material on MOSFET power amplifiers out there in books.
Cheers,
Bob
Hi Pete,
That's interesting about the speed of the 40409/10 devices. I really did not realize they were that fast. I think there were several vintages of the TO3 version of the 2N3055, with different ft numbers ranging from possibly below 1 MHz to around 2.5 MHz.
I built two clones of the Citation 12 back in 1970.
Cheers,
Bob
Hello Bob,
I am getting your book! I love good books.
Warm Regards,
James
Hi James,
I appreciate your interest in my book and really hope you enjoy it. I'll definitely be interested in reading your opinions after you read it.
Cheers,
Bob
I have actually seen some designs (and actual amps) that use the 340/350 do just that ..
OS
Good point.
Using Ccb alone for Miller compensation would be bad for both stability margin and distortion, due to its nonlinearity. Designs which precede the VAS transistor with an emitter follower largely kill the influence of nonlinear Ccb.
Cheers,
Bob
Hi Pete,
This is a really neat circuit technique. I wonder if there are any gotcha's? Looks like its set up for 2.5 mA. At lower frequencies, like 10 MHz, the 0.01 uF cap might want to go to 0.1. Using this circuit, we should be mindful that we oparate at a frequency where ac current gain is substantially smaller than LF current gain and preferably moderately greater than unity.
Cheers,
Bob